基于超复数奇异值分解的滚动轴承多通道振动数据压缩

针对滚动轴承的多通道振动信号,提出一种基于超复数奇异值分解(HC-SVD)压缩方法.在振动试验中,应用多通道传感器获取滚动轴承三正交轴x、y、z的振动数据,将三通道振动数据构成一纯四元数序列,并应用超复数奇异值分解法提取振动数据的超复数主元成分进行压缩.实验结果表明:该方法能剔除冗余信息,具有很好的重构性,而且能保持在采样时刻上三轴信号间的振动机理关联性,为多通道信号压缩与特征分析提供了一种新方法.

An Approach for Data Compressing of Multichannel Vibration Signals Based on Hyper-complex Number SVD for Rolling Bearings

A method of compressing multichannel vibration signals of rolling bearings was proposed based on hyper-complex SVD(HC-SVD).In vibration experiments,the vibration signals of x,y,and z orthogonal axis of rolling bearing were gotten by a multi-channel sensor,and they were used to construct a pure quaternion matrix to be the sample for HC-SVD processing.The hyper-complex principal components were extracted by it,and the orthogonal axis vibration signals were compressed used the extracted principal components.Th...